This study attempts to analyze patterns of intra-sentential code-switching in the speech of Russian children placed in the English-speaking environment and to characterize structural and procedural options which they prefer for forming mixed words. Code-switching at the word level is viewed here as a symptom of first language attrition. Previous studies of code-switching in children (see Romaine 1989 for summary) and attrition of Russian (e.g., Turian and Altenberg 1991, Polinsky, in press ) did rarely focus on morphological aspect of attrition.
The data consists of transcriptions of personal interviews, occasional tape-recordings of conversations, and notes of parents of 12 Russian-born children between the ages of seven and twelve who have been living in an L2 setting for more than three years. Despite their parents' attempts to maintain L1 at home, availability of Russian cultural events and frequent visits of monolingual Russian relatives and friends, immersion into the English dominant environment has lead to decay of certain lexical and grammatical forms which is revealed in code-switching at the word level. Structural differences between inflectional Russian and analytical English highlight features of word-formation processes in bilingual children. An investigation into formal aspects of the children's mixing reveals structural consistency they maintain in their code-switching: L1 inflectional and derivational morphemes are attached to L2 nominal, verbal and adjectival stems in 98% of cases. Only in 7 cases (2%) the L2 plural morpheme -s and present progressive -ing were used with L1 stems. Mixed verbs (50%) and nouns (38%) occur more often than adjectives (10%) in the data. Children in study employ suffixation, prefixation, postfixation, and the combination of these processes to express the intended meaning. Their repertoire of word-formation processes is similar to those of Standard Russian but the scope of morphemes is limited to highly productive L1 affixes. The main mechanism responsible for producing mixed forms in the children's speech is analogy where an L1 word or a pattern of form- or word-formation serves as a model. It is hoped that the findings of this study can provide further insights into the internal operations of language in a language-contact situation. In future it would be interesting to compare the children's and adults' patterns of Russian-English code-switching at a word level to make the issues of larger attrition process clearer.